SampleParaCal

A comprehensive Python package for calculating sample sizes in medical diagnostic studies. This package implements a wide range of statistical methods for diagnostic research, including:

• Sample size calculations for diagnostic accuracy studies
• ROC curve analysis
• Comparison of multiple diagnostic tests
• Non-inferiority and equivalence testing
• Multi-reader multi-case (MRMC) study design

Installation

pip install spcal

Quick Start

import spcal as spc

# Calculate sample size for a study using a two-sided confidence interval
n = spc.two_sided_CI_sample_size(var=0.25, alpha=0.05, L=0.1)
print(f"Required sample size: {n}")

# Calculate AUC variance
variance = spc.AUC_variance_binormal(A=0.85, R=1.5)
print(f"AUC variance: {variance:.6f}")

# Compare two diagnostic tests
n = spc.sample_size_for_two_diagnostic_tests(
    alpha=0.05, 
    beta=0.2, 
    delta=0.1, 
    Se1=0.85, 
    Se2=0.75, 
    coPos=0.6
)
print(f"Sample size for comparison: {n}")

Features

Single Diagnostic Method Evaluation

• Confidence interval-based sample size calculations
• Area Under ROC Curve (AUC) variance estimation
• Partial AUC analysis
• High-accuracy test evaluation
• Clustered data analysis

Threshold Optimization

• Sensitivity calculation at fixed false positive rates
• Variance estimation for transformed sensitivity
• Binormal ROC curve modeling

Diagnostic Method Comparison

• Paired and unpaired sample size calculations
• Relative sensitivity and specificity comparisons
• Covariance estimation for correlated tests
• Predictive value comparison (PPV/NPV)

Non-inferiority and Equivalence Testing

• Non-inferiority sample size calculations
• Equivalence testing for diagnostic methods
• Clustered data equivalence testing

Multi-reader Studies

• Variance components for reader variability
• Sample size for multi-reader studies
• Multi-reader multi-case (MRMC) study design

License

This project is licensed under the MIT License - see the LICENSE file for details.

Demo Book Reference